Full proposal available here: https://www.cosmostat.org/jobs/unbiased-shear-estimation-for-euclid-with-automatically-differentiable-and-gpu-accelerated-modeling

Weak gravitational lensing, the apparent distortion of the shapes of galaxies caused by matter along the line of sight, is a powerful method for probing the nature of dark matter, dark energy, and gravity. Following the launch of the Euclid satellite in 2023, we are entering a new era of precision cosmology. The instruments onboard Euclid have been designed to measure weak-lensing signals across a large fraction of the sky to unprecedented levels. The success of this mission, however, is predicated on the ability to measure the degree to which galaxy images are distorted — an anisotropic distortion called shear.

Traditionally, and as is currently implemented within the Euclid pipeline, the shear field is probed by measuring the shapes of galaxy images and examining how these shapes are correlated on different scales. However, all shape-estimation methods introduce a bias into the measurement of shear, as the modelling does not account for the full morphological diversity of galaxies in the Universe. While significant effort has been invested in recent years in methods to calibrate for this bias, another approach would be to investigate unbiased estimators of shear.

One such technique, pioneered within the CosmoStat team (Rémy et al. 2022; Centofanti et al., in prep.; Ayçoberry et al., in prep.), has been to build a forward model of the problem in order to directly infer the shear field. This involves simulating galaxy images using state-of-the-art deep-learning architectures, applying a known shear, and comparing the results with real observations. This forward-modelling approach has demonstrated very compelling results on simulated data.

The successful PhD applicant will work on extending this forward-modelling framework to real Euclid data. This will involve updating the pipeline to account for the full complexity of the Euclid Science Ground Segment (SGS) and ensuring the robust scalability of the technique to handle ~14 000 deg² of sky. The student will be able to build upon existing efforts within the Euclid Consortium to develop forward-modelling tools, as well as to directly compare with the shape-measurement methods implemented within the SGS.

This topic is particularly well timed with the first public release of Euclid data expected in October 2026. This will provide the student with an active and dynamic environment of researchers aiming to quantify the impact of measured shear bias on cosmological parameter inference. Any advancement towards a truly unbiased shear estimate could therefore have a significant impact on the field.

=================

For more Information about the topics and the co-financial partner (found by the lab!); contact Directeur de thèse - samuel.farrens@cea.fr

Then, prepare a resume, a recent transcript and a reference letter from your M2 supervisor/ engineering school director and you will be ready to apply online  before March 13th, 2026 Midnight Paris time!